Association of Catsper1 or -2 with Cav3.3 Leads to Suppression of T-type Calcium Channel Activity*

Sperm-specific CatSper1 and CatSper2 proteins are critical to sperm-hyperactivated motility and male fertility. Although architecturally resembling voltage-gated ion channels, neither CatSper1 nor CatSper2 alone forms functional ion channels in heterologous expression systems, which may be related to the absence of yet unidentified accessory subunits. Here we isolated CatSper1- and CatSper2-associated protein(s) from human sperm and analyzed their identities by a multidimensional protein identification technology approach. We identified the T-type voltage-gated calcium channel Cav3.3 as binding to both CatSper1 and CatSper2. The specificity of their interactions was verified by co-immunoprecipitation in transfected mammalian cells. Electrophysiological studies revealed that the co-expression of CatSper1 or CatSper2 specifically inhibited the amplitude of Cav3.3-evoked T-type calcium current without altering other biophysical properties of Cav3.3. Immunostaining studies revealed co-localization of CatSper1 and Cav3.3 on the principal piece of human sperm tail. Furthermore, fluorescence resonance energy transfer analysis revealed close proximity and physical association of these two proteins on the sperm tail. These studies demonstrate that CatSper1 and CatSper2 can associate with and modulate the function of the Cav3.3 channel, which might be important in the regulation of sperm function.

Sperm must swim long distances in the female reproductive tract to reach the site of fertilization. In addition, penetration through the gelatinous zona pellucida layer of the oocyte requires the sperm to swim in a hyperactivated state at the time and site of fertilization. This process is characterized by high amplitude and asymmetric beating of the sperm tail. Many studies have indicated that Ca 2ϩ serves as a key regulator in the initiation and maintenance of motility, including the hyperactivated motility (for reviews see Refs. 1 and 2).
Extensive efforts have been undertaken to identify and characterize Ca 2ϩ entry pathways, particularly Ca 2ϩ channels, involved in sperm motility processes. Two such ion channellike proteins, CatSper1 and CatSper2, were shown to be specifically expressed in the principal piece of the sperm tail (3,4).
Targeted disruption of CatSper1 led to sterile phenotypes in otherwise normal male mice. Further studies revealed that mutant sperm lack hyperactivated motility (5). In vitro fertilization assays revealed that CatSper1 mutant sperm could not fertilize eggs with an intact zona pellucida layer but could fertilize eggs whose outer layers had been enzymatically removed (3). Targeted disruption of CatSper2 also led to male sterile phenotypes, and the null sperm has identical loss-of-function phenotypes as does the CatSper1 null sperm (6,7). Therefore, CatSper1 and CatSper2 proteins appear to be essential for hyperactivated motility needed late for the sperm to penetrate the zona pellucida.
CatSper1 and CatSper2 represent a unique class of putative ion channel proteins (for a review see Ref. 8). They contain a single domain comprised of six transmembrane-spanning regions, akin to the voltage-gated potassium channels (3). However, their ion selectivity pore sequences between transmembrane regions 5 and 6 are closest to a single domain of the much larger voltage-gated Ca 2ϩ -selective channels. Residues lining the fourth transmembrane region of CatSper resemble a voltage sensor, as described for voltage-gated ion channels. However, recording ion channel activity following expression of CatSper subunits in heterologous expression systems, including Xenopus oocytes, HEK, 2 and Chinese hamster ovary K1 cells have not been successful (3,4). Attempts to measure whole cell currents from the sperm have proven difficult until very recently, where an alkaline-activated Ca 2ϩ current was recorded from sperm by patch clamp measurements (9). Interestingly, this current is absent in sperm lacking CatSper1. However, considering the co-dependent expression of CatSper1 and CatSper2 (7), it is still not clear which subunit or subunit complexes mediate this current.
Although several possibilities may be envisaged, the inability to functionally express CatSper subunits may be due to the following: (i) the absence of sperm-specific accessory proteins necessary for a putative ion channel complex in the heterologous systems, as observed with other ion channel complexes, or (ii) that CatSper may function as an accessory subunit to modulate the function of the principal subunit of an undetermined ion channel. The potential for either scenario exists, and it is indeed noteworthy that coiled-coil protein-protein interaction * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed: Neuroscience Research, domains are present in the C-terminal regions of each of the CatSper subunits (10). In this study, we adopted a GST pulldown approach to isolate CatSper-associated proteins from human sperm extracts, and we have identified the T-type calcium channel subunit Ca v 3.3 as an interacting protein. Electrophysiological studies revealed that the co-expression of either CatSper1 or CatSper2 specifically inhibited the amplitude of Ca v 3.3-evoked T-type Ca 2ϩ current without altering other biophysical properties of Ca v 3.3. Considering that CatSper1 and Ca v 3.3 subunits are co-expressed and associated with each other on the tail of the human sperm, our observations suggest that CatSper-Ca v 3.3 interactions could play an important role in regulating sperm functions such as hyperactivated motility.

EXPERIMENTAL PROCEDURES
Cloning-To generate FLAG-tagged human CatSper1 (GenBank TM accession number AF407333) and CatSper2 (GenBank TM accession number AF411818) expression constructs, the coding sequences of CatSper with an in-frame FLAG tag sequence were amplified by PCR and cloned into the NotI site of pIRESneo3 (Clontech). To generate pGEX-4T-1-Cat1-C to express the C terminus of CatSper1 as a GST fusion protein, the coding sequence covering the C-terminal 111 amino acids of human CatSper1 was amplified by PCR and cloned in-frame with GST in pGEX-4T-1 vector (Amersham Biosciences). To prepare pGEX-4T-1-Cat2-N and pGEX-4T-1-Cat2-C constructs for the expression of the N and C termini of CatSper2 as GST fusion proteins, respectively, the coding sequences covering the N-terminal 103 amino acids and C-terminal 190 amino acids of human CatSper2 were amplified by PCR and cloned in-frame with GST in pGEX-4T-1.
To clone human Ca v 3.3 (GenBank TM accession number AF393329), portions of the coding sequence were amplified by reverse transcription-PCR against poly(A) ϩ RNA from human cerebellum (Clontech). These fragments were ligated together and cloned in pcDNA3.1/V5-His TOPO expression vector (Invitrogen) so as to generate pcDNA3.1-Ca v 3.3-V5/His with V5 and His tags in-frame with the C terminus of Ca v 3.3.
Antibodies-To generate antibody against human CatSper1, the C-terminal 111 amino acids of CatSper1 as a GST fusion protein were purified from bacteria and were used to immunize rabbits. The CatSper1 C-terminal fragment, which was released from the GST fusion protein by thrombin cleavage, was used to affinity-purify antibodies against the CatSper1 part from the crude serum. The strategy as described by Quill et al. (4) was adopted to generate rabbit antibody against a peptide derived from the C-terminal 27 amino acids of mouse CatSper2. The resulting antibody recognized the human CatSper2 as well. The following antibodies were obtained from the sources indicated: human Ca v 3.3 antibodies (Santa Cruz Biotechnology, Santa Cruz, CA, and Alomone Labs, Israel), Na ϩ /K ϩ -ATPase antibody (Abcam, Cambridge, UK), PMCA4 antibody (Sigma), human 14-3-3⑀ antibody (Assay Designs, Ann Arbor, MI), V5 tag antibody (Invitrogen), and FLAG tag antibody (Sigma).
To prepare human sperm extract, cryo-preserved sperm samples were washed to remove the seminal fluid and lysed in lysis buffer (50 mM Tris, pH 7.5, 250 mM NaCl, 1% Nonidet P-40, 1% sodium deoxycholate, 2 mM EDTA, 0.5 mM DTT and protease inhibitor mixture from Sigma) for 1 h. The extract was centrifuged to remove cell debris and combined with an equal volume of dilution buffer to bring down Nonidet P-40 and sodium deoxycholate concentration to 0.5%. The sperm extract was precleared by incubation with glutathione-Sepharose 4B beads for 1 h. To pull down CatSper-associated proteins, the sperm extract was divided into equal parts, and beads conjugated with GST (as a negative control), GST-Cat1-C, or a mixture of GST-Cat2-N and GST-Cat2-C proteins were added. After incubation at 4°C with gentle rotation overnight, the beads were washed three times with 10 ml of ice-cold IP buffer (50 mM Tris, pH 7.5, 250 mM NaCl, 0.5% Nonidet P-40, 0.5% sodium deoxycholate, 2 mM EDTA, 0.5 mM DTT and protease inhibitor mixture from Sigma). The GST fusion proteins along with associated proteins on the washed beads were eluted in a buffer containing glutathione (100 mM Tris, pH 8.0, 150 mM NaCl, 20 mM glutathione, 0.2% Triton X-100).
MudPIT Analysis of Protein Identities-Half of the eluted protein mixtures from GST pulldown experiments were resolved by SDS-PAGE and visualized by silver staining. Selective protein bands on the gel were excised, and the gel slices were destained and chopped into 1-mm size cubes. A slightly modified procedure originally developed by Shevchenko et al. (11) was employed for in-gel digestion. The extracted peptides were lyophilized and resuspended in 15-20 l of 5% formic acid until further analysis. The other portion of the eluted protein mixtures were subjected to in-solution digestion (12) resulting in complex peptide mixtures. Peptide digests resulting from in-gel digests or in-solution digests were then individually loaded on to a three-phase MudPIT (RP-SCX-RP) column. A three-step MudPIT analysis was used for analyzing the in-gel digests, and a six-step MudPIT analysis was used for in-solution digests (13).
Transfection and Cell Culture-HEK cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin in a humidified 5% CO 2 , 95% O 2 incubator at 37°C. Cells were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. To generate a cell line stably expressing Ca v 3.3, pcDNA3.1-Ca v 3.3-V5/His was linearized by SalI and transfected into HEK cells. Stable cell line was generated by antibiotic selection (1 mg/ml G418) 48 h posttransfection. Single cell colonies were selected 14 days posttransfection and amplified, and the expression of Ca v 3.3 was assessed.
Co-immunoprecipitation-HEK cells in 10-cm dishes were transiently transfected with equal amounts of the expression constructs (totally 10 g of DNA) using Lipofectamine 2000. When only a single construct was transfected, pcDNA3.1 vector alone was included to maintain the same final amount of DNA. Cells were lysed 48 h post-transfection with IP buffer (50 mM Tris, pH 7.5, 250 mM NaCl, 0.5% Nonidet P-40, 0.5% sodium deoxycholate, 2 mM EDTA, 0.5 mM DTT, and protease inhibitor mixture from Sigma). The cell lysates were then centrifuged to remove cell debris, and the antibody used for immunoprecipitation was added to the supernatant. After overnight incubation at 4°C with gentle agitation, immune complexes were precipitated with protein A-or protein G-agarose beads (Invitrogen) followed by washes in 1 ml of IP buffer for three times. After the final wash, the pellet was resuspended in Laemmli sample buffer, and proteins were resolved by SDS-PAGE (4 -12% gel) and transferred to polyvinylidene fluoride membrane for immunoblot analysis.
Whole Cell Patch Clamp-HEK cells stably expressing Ca v 3.3 were transfected with either pcDNA3.1 vector, CatSper1, or CatSper2 expressing constructs using Lipofectamine 2000 (Invitrogen). Cells were also co-transfected with a plasmid encoding a GFP reporter (in a 1:5 ratio) to allow identification of positively transfected cells for whole cell patch clamp measurements. 24 -48 h post-transfection, whole cell currents were recorded at room temperature using the standard patch clamp technique with an Axopatch 200B amplifier (Axon Instruments, Union City, CA), controlled with a PC computer using pCLAMP6 software (Axon Instruments). Data were filtered at 5 kHz using the built-in filter of the amplifier. Borosilicate pipettes with a typical resistance of 2-3 megohms were filled with a solution containing the following: 110 mM CsCl, 10 mM EGTA, 10 mM HEPES, 2 mM MgATP, 0.6 mM GTP (pH adjusted to 7.2 with CsOH). Extracellular solution contained 5 mM CaCl 2 , 155 mM tetraethylammonium chloride, 10 mM HEPES (pH adjusted to 7.4 with tetraethylammonium-OH).
Data Analysis-In electrophysiology measurements, peak currents were determined using Clampfit 8.0 software (Axon Instruments). The conductance-voltage relationship for activation was deduced by the chord conductance method, wherein conductance was obtained by normalizing peak current at each pulse against driving force, plotted as function of voltage, fit with a single Boltzmann function, and normalized against maximal value. Average data are presented as mean Ϯ S.E., and statistical differences between data sets were assessed by Student's t tests, and significance was accepted at the p Ͻ 0.05 level.
Biotinylation of Cell Surface Proteins for Expression Analysis-Cells were transfected with a fixed amount of DNA (10 g) in 10-cm dishes using Lipofectamine 2000. 0.5 g of a plasmid expressing an hemagglutinin-tagged protein was included in each transfection to allow normalization of transfection efficiencies among various samples. 48 h post-transfection, cells were harvested, washed with PBS, and then incubated in PBS containing 0.5 mg/ml sulfo-NHS-LC-biotin (Pierce) for 45 min at 4°C to label cell surface proteins. The biotinylation reaction was quenched with 50 mM NH 4 Cl for 10 min. The cells were washed with PBS and incubated in lysis buffer (0.5% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM DTT, protease inhibitor mixture in PBS). The cell lysates were then centrifuged to remove cell debris, and proteins in the supernatant were quan-titated. Equal amounts of protein lysates were aliquoted, one part for the measurement of total protein expression, and the rest were incubated with beads coated with streptavidin (Pierce) for affinity-capture and purification of biotinylated proteins. The proteins were resolved by SDS-PAGE (4 -12% gel) and transferred to polyvinylidene fluoride membrane for immunoblot analysis. The intensities of protein bands within a linear range of detection on the scanned blot were quantitated by ImageQuant image analysis software, and the expression levels of protein were normalized to that of the transfection control protein.
Immunofluorescence-Cryo-preserved human sperm samples were thawed, diluted, spotted onto chamber slides, and air-dried. The sperm were fixed and permeabilized with 4% paraformaldehyde for 10 min. After rinsing twice in PBS, the slides were incubated in blocking buffer (2% fetal bovine serum, 2% bovine serum albumin in PBS) for 30 min to reduce nonspecific binding. The slides were then incubated with either rabbit anti-Ca v 3.3 antibody (Alomone Labs, 1:50 dilution) or rabbit anti-CatSper1 antibody (1:50 dilution) diluted in the blocking buffer for 2 h. To assess specific binding, 2 g of competing peptide for anti-Ca v 3.3 antibody or 25 g of purified CatSper1 C-terminal fragment for competing anti-CatSper1 antibody were preincubated with the primary antibodies in the blocking buffer for 30 min and then applied to the slides. After incubating with primary antibodies, the slides were washed once in PBS with 0.25% Nonidet P-40 and then twice in PBS. The slides were then incubated with fluorescein isothiocyanate-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch, West Grove, PA) diluted in blocking buffer (1:200 dilution) for 1 h. After washing one time in PBS with 0.25% Nonidet P-40 followed by two times in PBS, the slides were mounted in Vectashield mounting medium (Vector Laboratories, Burlingame, CA) for observation under a fluorescent microscope.
FRET Analysis-Anti-Ca v 3.3, anti-CatSper1, and anti-PMCA4 antibodies were labeled with Alexa Fluor 488 dye or Alexa 555 dye by Zenon IgG labeling kit (Invitrogen) according to the manufacturer's protocol. These labeled antibodies were employed to immunostain sperm samples, as described for indirect immunofluorescence, but without using the secondary antibody. Before mounting, the stained sperm on the slide were fixed again with 4% paraformaldehyde for 10 min. Fluorescent images were acquired using an LSM 5 PASCAL laser scanning confocal imaging system (Carl Zeiss, Thornwood, NY). All images were taken with an oil immersion objective with appropriate filter sets (donor, excitation 488 nm, emission filter BP 505-530 nm; acceptor, excitation 543 nm, emission filter LP 560 nm; FRET, excitation 488 nm, emission filter LP 560 nm). The quantitative FRET analysis was performed using the PAS-CAL software (Zeiss) according to manufacturer's sensitized emission protocol, which is based on the conventional threefilter method described by Xia and Liu (14). The FRET filter raw image contains the FRET signal, as well as the bleed through of direct donor and acceptor emissions into the FRET channel. To determine bleed through and background corrections, the three track images of sperms stained with donor only and acceptor only were separately obtained. The donor and acceptor images of the samples were then multiplied with the respec-tive correction factor and subtracted from the raw FRET image, and a normalized FRET (NFRET) image was calculated by PAS-CAL software according to Xia and Liu (14). NFRET values of at least 10 regions on the tails of 3-5 stained sperm were measured and expressed as means Ϯ S.E.

Identification of Ca v 3.3 as a CatSper1-and CatSper2-associated Protein from Human Sperm Extracts by GST Pulldown
Approach-To identify sperm proteins that associate with CatSper1, we initially tried to immunoprecipitate CatSper1 along with its binding partners from the sperm extract using CatSper1 antibody. However, this approach failed because of the poor solubility of sperm CatSper1 protein under conditions that maintain protein-protein interactions. Because CatSper proteins have relatively large N-and C-terminal segments and short loops between the transmembrane domains, we hypothesized that CatSper might associate with other proteins through their N-and C-terminal segments. In particular, the C-terminal segments of both CatSper1 and CatSper2 contain coiled-coil domains that could mediate protein-protein interactions (10). Therefore, we generated the N-and C-terminal segments of CatSper as GST fusion proteins and utilized them in pulldown experiments to identify proteins in sperm that may associate with them.
GST fusion protein constructs were prepared and expressed in bacteria. The N terminus of CatSper1 as a GST fusion protein expressed poorly and was largely insoluble, whereas good expression of the C terminus of CatSper1 and the N and C termini of CatSper2 as GST fusion proteins (GST-Cat1-C, GST-Cat2-N, and GST-Cat2-C, respectively) were obtained. These GST fusion proteins were affinity-purified to apparent homogeneity and conjugated to glutathione-Sepharose 4B beads. To pull down CatSper-associated proteins, human sperm extract was incubated with beads conjugated with GST (negative control), GST-Cat1-C, or a mixture of GST-Cat2-N and GST-Cat2-C proteins, respectively. Aliquots of the pulldown products were resolved by SDS-PAGE and visualized by silver staining. GST fusion proteins used in pulldown appeared as major protein bands on the gel. Besides, less abundant protein bands appeared in all three lanes of Fig. 1. Some of these protein bands, with identical molecular weights across all lanes, were apparent nonspecific contaminants during purification. Nonetheless, distinct protein bands were observed specifically in GST-CatSper1 and -2 pulldown lanes but not in GST pulldown controls.
We applied MudPIT, which incorporates on-line two-dimensional capillary chromatography coupled to tandem mass spectrometry to determine the identities of proteins in the pulldown mixtures. As expected, several peptides derived from either GST or CatSper were identified because of the presence of the GST fusion proteins in the pulldown mixtures. To identify proteins specifically associated with CatSper, we subtracted, in silico, proteins identified in GST alone pulldown from those obtained from GST-CatSper1 and -2 pulldowns. This subtractive analysis identified a few proteins present/ unique to the GST-Cat1-C or a mixture of GST-Cat2-N and GST-Cat2-C pulldown. Some identified proteins were nonspe-cific contaminants from semen, including apolipoprotein and semenogelin. One peptide (RTFRLLRVLKLVRFMPALRR) derived from the T-type calcium channel Ca v 3.3 was consistently identified (from two separate experiments) in both in-gel and in-solution digestion of GST-CatSper1 and GST-CatSper2 pulldown complexes but not from pulldown product by GST alone. These findings suggested that Ca v 3.3 might be a likely associated protein for both CatSper1 and CatSper2.
Co-immunoprecipitation of CatSper1 and CatSper2 with Ca v 3.3 in Mammalian Cells-In order to confirm the association of Ca v 3.3 with the CatSper1 and -2 observed in GST pulldown, co-immunoprecipitation experiments were conducted to investigate their interactions in mammalian cells. CatSper1 with a FLAG tag was transiently transfected into HEK cells either alone (negative control) or along with a construct expressing Ca v 3.3 with a V5 tag. Forty eight hours post-transfection, the expression of transfected Ca v 3.3 and CatSper1 could be detected by Western blot analysis of products immunoprecipitated with the corresponding antibodies ( Fig. 2A). When Ca v 3.3 antibody was used in the immunoprecipitation, we observed that CatSper1 co-immunoprecipitated with Ca v 3.3 in the case when both were co-transfected but not when CatSper1 alone was transfected ( Fig. 2A). To determine the specificity of interactions between Ca v 3.3 and CatSper1, we examined whether other unrelated proteins, such as Na ϩ /K ϩ -ATPase and 14-3-3⑀, could also co-immunoprecipitate with Ca v 3.3. Although these two proteins were abundantly expressed, they were not present in the products immunoprecipitated by the Ca v 3.3 antibody ( Fig. 2A). Similar co-immunoprecipitation studies revealed that CatSper2 also associates with Ca v 3.3 in mammalian cells (Fig. 2B). In reciprocal experiments, when antibodies recognizing the cloned CatSper proteins were used in immunoprecipitation, we observed that Ca v 3.3 could be co-immunoprecipitated along with either CatSper1 or CatSper2 (Fig. 2C).

CatSper1 and CatSper2 Reduce Ca v 3.3-mediated Ca 2ϩ Currents in Transfected Mammalian Cells and Xenopus
Oocytes-The physical association of Ca v 3.3 with CatSper1 and -2 raised the possibility of functional interactions between these proteins. We generated HEK cells stably expressing CatSper1 and CatSper2, and in consistent with earlier reports (3,4), neither CatSper1 nor CatSper2 alone elicited functional responses as measured by whole cell patch clamp studies (data not shown). We next generated a cell line stably expressing Ca v 3.3 and transiently transfected either a pcDNA3.1 vector (control) or CatSper1 or -2 expression constructs, along with a GFP reporter construct. Whole cell patch clamp experiments were performed on selected GFP-positive cells. By using 300-ms voltage pulses ranging from Ϫ100 to ϩ50 mV in 5-mV increments from a holding potential of Ϫ110 mV, we observed voltage-dependent current responses in cells expressing Ca v 3.3 alone or in combination with CatSper1 or CatSper2. Representative current traces are shown in Fig.  3A. These traces displayed typical features of T-type currents such as activation at low voltages and crossing over of current traces at certain potentials (15). Interestingly, co-expression CatSper1 or -2 with Ca v 3.3 significantly reduced current amplitudes as evidenced by the currentvoltage (I-V) relationship (Fig. 3B). The maximal current densities were reduced significantly from a control (Ca v 3.3 alone) density of Ϫ124.9 Ϯ 10.1 pA/pF to Ϫ66.5 Ϯ 9.8 pA/pF (46.7%) and Ϫ75.4 Ϯ 4.4 pA/pF (39.6%) for Ca v 3.3/Catsper1 and Ca v 3.3/CatSper2, respectively. To examine the effect on the activation parameters, the conductance-voltage relationship for activation was derived from current-voltage relationship curves (Fig. 3C). The V1 ⁄ 2 and slope factors for Ca v 3.3 were Ϫ44.5 Ϯ 0.5 and 7.8 Ϯ 0.5 mV (n ϭ 9), respectively. The V1 ⁄ 2 values were slightly shifted by ϩ1.3 and ϩ2.9 mV for Ca v 3.3/CatSper1 and Ca v 3.3/CatSper2 combinations, respectively (p Ͻ 0.1).
The voltage dependence of channel availability was determined by a two-pulse protocol (Fig. 4A). From a holding potential of Ϫ110 mV, a 3-s pulse ranging from Ϫ110 to Ϫ20 mV was applied to allow channel inactivation, and a second pulse to Ϫ30 mV was applied to assess relative channel availability. Peak currents elicited by second pulse were plotted as function of voltage of the first pulse and fit with Boltzmann function to determine the maximal peak currents. The normalized currents against maximal peak currents were plotted and fit with Boltzmann function to derive voltage-dependent inactivation. Neither the V1 ⁄ 2 nor slope factors were significantly altered by co-expression with CatSper1 or -2 (Fig. 4B).
We also investigated the effects of co-expression of CatSper1 on the electrophysiological properties of Ca v 3.3 in another expression system, viz. Xenopus oocytes. Current responses were measured following co-injection of Ca v 3.3 and CatSper1 cRNA (90 ng each). Similar to effects observed in HEK cells, co-injection of CatSper1 reduced peak currents at Ϫ25 mV by 42 Ϯ 6% (n ϭ 22), whereas other biophysical properties of Ca v 3.3 were largely unaffected (data not shown). To exclude the possibility that the expression of CatSper1 might nonspecifically affect other channels besides Ca v 3.3, we co-expressed CatSper1 with HCN2, which encodes the hyperpolarizationactivated cyclic nucleotide-gated channel. Overexpression of CatSper1 does not affect the expression level of HCN2 (data not shown), supporting the idea that the inhibition of Ca v 3.3 currents is because of its specific interaction with CatSper1.

Effect of Co-expression of CatSper1 on the Surface and Total Expression Levels of Ca v 3.3 in Transfected
Cells-A possible explanation for the effects of CatSper expression on the ampli-tude of the Ca v 3.3-evoked current is that CatSper proteins might affect the expression level of Ca v 3.3 on the cell surface. To explore this possibility, Ca v 3.3, either with or without CatSper1, along with a control plasmid (to normalize transfection efficiencies among different samples) were transfected into cells. Forty eight hours post-transfection, surface membrane proteins of viable, intact cells were labeled with an impermeable biotin probe followed by affinity purification of the biotinylated proteins using the streptavidin beads. The detection of plasma membrane marker Na ϩ /K ϩ -ATPase, but not the cytoplasmic protein 14-3-3⑀, in the biotinylated protein fraction validated the use of biotinylation technique for assessing cell surface expression of proteins (Fig.  5A). The expression levels of Ca v 3.3 in the biotinylated cell surface fractions and in total cells were examined by Western blot analysis (Fig.  5A) and then quantitated and normalized against the expression levels of a transfection control protein (Fig. 5B). The normalized surface and total expression of Ca v 3.3 in the absence of CatSper1 were 75 Ϯ 10 and 138 Ϯ 12, respectively, whereas in the presence of CatSper1, the normalized surface and total expression were reduced to 61 Ϯ 12 and 119 Ϯ 15, respectively (n ϭ 3; Fig. 5B). Thus, a modest reduction in the levels of surface or total expression of Ca v 3.3 was observed when co-expressed with CatSper1 (although this did not achieve statistical significance p Ͻ 0.05).

Co-expression of CatSper1 and Ca v 3.3 on the Principal Piece of Human Sperm Tail-To investigate whether CatSper1 and
Ca v 3.3 proteins are co-expressed in human sperm, an indirect immunofluorescence technique was used. Unlike the observations by Trevino et al. (16), we could not conclusively determine Ca v 3.3 expression using the same anti-Ca v 3.3 antibody (purchased from Santa Cruz Biotechnology) because immunostaining was only partially blocked by the corresponding antigen. We instead relied on an anti-Ca v 3.3 antibody from a different source (Alomone Labs) in our immunostaining experiments (17). Western blot analysis revealed that this antibody not only recognize cloned Ca v 3.3 expressed in HEK cells but also detected Ca v 3.3 expressed in human sperm and cerebral cortex (Fig. 6A). When this anti-Ca v 3.3 antibody was used in immunostaining experiments, robust staining on the principal piece and a slightly weaker staining on the middle piece of the sperm tail was observed (Fig. 6B, panel 2). The staining was totally blocked by preincubation of the primary antibody with the corresponding antigen peptide (Fig. 6B, panel 4). We next examined the expression of CatSper1 in human sperm. We observed robust fluorescence signals on the principal piece of Currents were elicited by voltage steps ranging from Ϫ100 to ϩ50 mV in 5-mV increments from a holding potential of Ϫ110 mV. B, current-voltage (I-V) relationships were obtained by measuring the peak currents during the 300-ms pulses and plotted as function of voltage (n ϭ 9, 6, and 6 for Ca v 3.3, Ca v 3.3/CatSper1, and Ca v 3.3/CatSper2, respectively). Currents peaked at Ϫ25 mV. C, the conductance-voltage relationship for activation was deduced by the chord conductance method, wherein conductance was obtained by normalizing peak current at each pulse against driving force, plotted as function of voltage, fit with a single Boltzmann function, and normalized against maximal value. The mean values were plotted and fitted with Boltzmann function to obtain V1 ⁄2 and slope factor k. Ca  sperm tail with the antibody raised against the C terminus of human CatSper1 (Fig. 6C, panel 2). This staining was blocked by preincubation of the primary antibody with the recombinant CatSper1 C-terminal fragment used to raise the antibody (Fig. 6C, panel 4), indicating specific staining of CatSper1 expression on the principal piece. This is consistent with the observation of CatSper1 localization on mouse sperm (3). The staining of both Ca v 3.3 and CatSper1 on human sperm tail demonstrates co-localization of these two proteins in this region.

FRET Analysis of CatSper1 and Ca v 3.3 Interaction on Human
Sperm Tail-We employed FRET microscopy to assess whether CatSper1 physically associates with Ca v 3.3 in sperm. FRET consists of radiationless energy transfer between one fluorophore (the donor) in the excited state and another fluorophore (the acceptor) when in close proximity. Simple co-localization of two proteins is not sufficient to yield energy transfer, which requires the proximity of the two fluorophores at distances that are on the order of a few nanometers. Thus, the presence of FRET between two proteins is an indicator of close proximity and protein-protein interaction (18,19).
We labeled anti-Ca v 3.3 antibody with Alexa Fluor 488 dye as the donor (Ca v 3.3-488) and anti-CatSper1 antibody with Alexa 555 dye as the acceptor (CatSper1-555). These two labeled antibodies were mixed to immunostain human sperm, and FRET analysis was performed using the sensitized emission method (14). A set of representative donors, acceptors, and raw FRET images are  shown in Fig. 7. An NFRET image, which shows FRET intensities with high spatial resolution, was calculated from the raw FRET image after correction for the effects of bleed through of direct donor and acceptor emissions into the FRET channel. We observed robust NFRET signals along the sperm tail but not in the head (Fig. 7), indicating a region-specific close molecular association between CatSper1 and Ca v 3.3. Quantitative measurement of 10 regions on the tails of different sperm samples revealed a mean NFRET value of 0.394 Ϯ 0.017. As a positive control, we labeled anti-CatSper1 antibody with either Alexa 488 dye as the donor or Alexa 555 dye as the acceptor and mixed them to immunostain human sperm (Fig. 7). Both labeled antibodies should be in close proximity by recognizing the same protein, and indeed as expected, we detected a robust NFRET signal (mean value of 0.416 Ϯ 0.012), which is comparable with that from the experimental group. To assess the specificity of the FRET analysis, we also immunostained sperm with anti-PMCA4 antibody labeled with Alexa 488 dye and anti-Ca v 3.3 antibody labeled with Alexa 555 dye (Fig. 7). Although PMCA4, a Ca 2ϩ -ATPase, is expressed in the sperm tail just like the CatSper1 and Ca v 3.3 (20), only marginal NFRET signals (mean value of 0.075 Ϯ 0.009) were observed in the stained sperm, indicating lack of interaction between Ca v 3.3 and PMCA4. Col-lectively, these results indicate that Ca v 3.3 specifically associates with CatSper1 on the sperm tail.

DISCUSSION
This study reports the identification and analysis of interactions of the voltage-gated T-type calcium channel Ca v 3.3 with putative sperm ion channel proteins CatSper1 and CatSper2. Initial evidence for such protein-protein interactions stems from MudPIT analysis of proteins from human sperm extracts that associate with the Catsper1 C terminus or a mixture of CatSper2 N and C terminus as GST fusion proteins. In both cases, a peptide derived from Ca v 3.3 was identified. Subsequent co-immunoprecipitation experiments confirmed the physical interactions between Ca v 3.3 and CatSper1 or CatSper2 in mammalian cells. Furthermore, our FRET analysis demonstrates that Ca v 3.3 and CatSper1 interactions do occur physiologically and suggests that these two proteins associate with each other on the tail of human sperm.
To our knowledge, Ca v 3.3 is the first protein identified that associates with CatSper1 and -2. The presence of coiledcoil domains on the C terminus of CatSper1 and CatSper2 suggests that the C-terminal segments of CatSper proteins likely mediate Ca v 3.3-CatSper interactions. However, the potential promiscuity of coiled-coil interactions with other proteins should be acknowledged, as it is not known whether other proteins with coiled-coil segments associate with Ca v 3.3 as well. The association of both CatSper1 and CatSper2 with Ca v 3.3 initially raised the possibility that these three proteins might form a triple complex. However, when we co-expressed all three proteins in mammalian cells, we failed to detect CatSper2 in the product immunoprecipitated by CatSper1 antibody or vice versa. 3 This implies that CatSper1 and CatSper2 may bind competitively to Ca v 3.3, perhaps through a region common in Ca v 3.3.
The physical association of Ca v 3.3 with CatSper1 and -2 suggested functional interactions between these proteins. However, we attempted, but failed, to elicit any currents other than the T-type calcium current from cells expressing both Ca v 3.3 and CatSper1 or -2, suggesting that co-expression of Ca v 3.3 still cannot facilitate the functional expression of CatSper in heterologous systems. It is likely, however, that Ca v 3.3 may not be the sole accessory protein that interacts with CatSper, and we could have missed the identification of other binding partners for CatSper1 or CatSper2 in our GST pulldown experiments. Other proteins in the sperm membrane might associate with CatSper proteins so tightly as a protein complex that precludes its capture by GST-CatSper in the pulldown procedure. We also noticed multiple distinct protein bands specifically in GST-CatSper pulldowns; however, we failed to resolve their identities. This might be due to dynamic range limitation of the Mud-PIT analysis, considering that the peptides derived from the GST fusion proteins are overwhelmingly abundant and may affect the sampling efficiency of the peptides from other relatively low abundant proteins.
Although co-expression of Ca v 3.3 with CatSper did not reconstitute a novel or distinct ion channel complex, we observed net reduction in Ca v 3.3-evoked T-type Ca 2ϩ currents upon co-expression with either CatSper1 or CatSper2 in heterologous expression systems. The whole cell current amplitudes of Ca v 3.3 showed significant reductions of 47 and 40% when co-expressed with CatSper1 and CatSper2, respectively. Interestingly, other biophysical properties of Ca v 3.3 were largely unaltered except for a modest (1-3 mV) shift in activation V1 ⁄2 of Ca v 3.3 toward the depolarizing direction. The effects of CatSper1 on Ca v 3.3 were replicated using the Xenopus oocyte expression system and are specific because overexpression of CatSper1 does not affect the expression of another channel, viz. HCN2. These studies demonstrate that the association of CatSper proteins with Ca v 3.3 predominantly affects the current amplitude of Ca v 3.3 without altering other channel properties.
Recombinant T-type Ca v 3 channel subunits generally do not require accessory proteins for functional expression in a variety of heterologous expression systems, which is in contrast to L-type calcium channels that are profoundly modulated by ␤ and ␣2␦ subunits and function as multimeric complexes. It has been reported, however, that ␤ 1b and ␣ 2 -␦ 1 subunits, typically associated with L-type calcium channels, can also modulate T-type calcium channels, including Ca v 3.3 (21). Unlike CatSper, these auxiliary subunits enhance the current amplitudes of Ca v 3 channels by increasing their cell surface expression. However, physical interactions between T-type Ca 2ϩ channel subunits and ␤ 1b or ␣ 2 -␦ 1 have not been established. In another report (22), it was shown that co-expression of calcium channel ␥ 6 subunit, but not the ␥ 4 or ␥ 7 subunits, with Ca v 3.1 in HEK cells significantly decreases current density without changing the kinetic properties and the protein expression of Ca v 3.1. To date, CatSper1 and CatSper2 remain as the first identified proteins shown to physically associate with and functionally modulate Ca v 3.3 channels. It would be interesting to determine whether CatSper proteins also modulate other T-type Ca 2ϩ channel types such as Ca v 3.1 and Ca v 3.2 that are also expressed in sperm.
How does the association of CatSper1 modulate Ca v 3.3 current? It is known that the macroscopic current is proportional to the product of the single channel conductance, the number of channels, the open probability, and the effective driving force. Because we did not see significant changes in the kinetic properties of Ca v 3.3, our data do not support the possibility that the decrease in current density is because of changes in Ca v 3.3 biophysical properties, although this cannot be entirely eliminated without detailed biophysical analysis at a single channel level. In a number of cases, it has been reported that auxiliary proteins may modulate the current density by altering the amount of channel protein on the cell surface (21,23). To assess whether the reduced amplitude could be attributed to reductions in the number of Ca v 3.3 channels, surface membrane proteins of viable, intact cells were labeled with biotin and quantified. We observed a tendency that co-expression of CatSper1 reduced the surface expression of Ca v 3.3 by about 20%, suggesting that reduction in expression of Ca v 3.3 might contribute to the decreased current density.
The low voltage-activated T-type Ca 2ϩ channels produce low threshold spikes that have been shown to trigger burst firing in various cell types. They play important physiological roles in diverse tissues, especially in central and peripheral nervous systems and in the heart (for a review see Ref. 24). In mammalian germ cells and sperm, the expression of all three members of the T-type Ca 2ϩ channels have been reported (16,25), and electrophysiological studies have documented the existence of T-type Ca 2ϩ current in mouse spermatogenic cells (26). The most prominent role attributed for T-type Ca 2ϩ channels is in the acrosome reaction. Blockade of T-type Ca 2ϩ channels during gamete interaction inhibited zona pellucidadependent Ca 2ϩ elevation and acrosome reaction (26). However, although T-type Ca 2ϩ channels are present in the sperm head (16), the expression of neither CatSper1 nor CatSper2 was localized in the acrosome region (3,4). Therefore, it is unlikely that Ca v 3.3-CatSper interactions could play a role in sperm acrosome reaction. Instead, the co-expression and association of Ca v 3.3 and CatSper1 on human sperm tail suggests a role for their interactions in sperm motility. Ca 2ϩ is a key regulator in the initiation and maintenance of hyperactivated motility of the sperm (27). During this process, Ca 2ϩ concentration in the cytoplasm rises, which regulates the movement of axoneme (28). The release of Ca 2ϩ from membrane-bound internal Ca 2ϩ stores, such as the redundant nuclear envelope, is critical for hyperactivated motility (29). Moreover, the increase of cytoplasmic Ca 2ϩ can also result from the influx of extracellular Ca 2ϩ through the plasma membrane; however, the type of Ca 2ϩ channels mediating this process remains elusive. Knock-out studies revealed that mice lacking either CatSper1 or CatSper2 have defects in sperm-hyperactivated motility that is accompanied by a reduction in Ca 2ϩ concentration in sperm tail (3,6), indicating a crucial role for these CatSper proteins in Ca 2ϩ influx during hyperactivated motility. The detection of Ca v 3.3 expression on sperm flagellum and, more importantly, its association with CatSper imply that Ca v 3.3 may also play a role in mediating calcium influx required for hyperactivated motility. However, studies with Ca v 3.3 knockout or with selective T-type Ca 2ϩ channel blockers are currently unavailable to directly assess the role of Ca v 3.3 in sperm function. It has been reported that weak T-type Ca 2ϩ channel inhibitors, mibefradil and gossypol, did not significantly affect sperm basal motility at low concentrations but did cause motility alterations at higher concentrations where high voltage-activated Ca 2ϩ channels may also be blocked (16). Accordingly, the assessment of the Ca v 3.3 role in sperm function awaits the identification of potent and selective Ca v 3.3 blockers or knock-out studies.
Should Ca v 3.3 indeed contribute to Ca 2ϩ influx during sperm-hyperactivated motility, CatSper1 could modulate Ca v 3.3 function by physical interactions in a region-specific manner. Interestingly, our studies demonstrate a decrease, rather than increase, in the amplitude of Ca v 3.3-mediated Ca 2ϩ currents upon co-expression of CatSper1 or CatSper2 in heterologous systems. Although it remains to be proven whether this holds true in the native sperm environment, this observation suggests that CatSper1 and -2 may "finetune" Ca v 3.3-mediated current to maintain an optimal level of Ca 2ϩ in the principal piece during hyperactivated motility. Indeed, recent studies reveal the presence of both major Ca 2ϩ influx and efflux mechanisms in the principal piece. PMCA4, which is a plasma membrane Ca 2ϩ -ATPase that acts as an extrusion pump to mediate the efflux of excess Ca 2ϩ from the cytosol, is highly expressed in the principal piece just like CatSper1 and Ca v 3.3. Targeted ablation of PMCA4 in mice resulted in defects in sperm-hyperactivated motility and male fertility (20,30). Abnormal mitochondria was observed in PMCA4 Ϫ/Ϫ sperm, which was attributed to Ca 2ϩ overload. Therefore, perhaps a controlled regulation of Ca 2ϩ levels, rather than mere increase, could be important for regulating hyperactivated motility of the sperm, and it may be speculated that Ca v 3.3 and CatSper proteins partner together to regulate Ca 2ϩ influx needed for hyperactivated motility.